1. Field of the Invention
The present invention relates to an exposure apparatus for optically transferring a circuit pattern, e.g., integrated circuit (IC) or the like, and more particularly the invention relates to an apparatus for detecting the characteristics of a focusing optical system which accurately aligns a mask or reticle with a wafer or an object to be exposed and forming an image of the mask or reticle on the object to be exposed.
Recently, there has been a rapid progress in the production of increasingly minute patterns of large-scale integrated circuits (LSI) and reduction projection type exposure apparatus have become popular as circuit pattern printing apparatus which satisfy the needs for the production of increasingly minute patterns and increase the productivity. The apparatus uses a mask or reticle (hereinafter typically referred to as a reticle) containing circuit pattern which is many times (e.g., ten times) larger than a circuit pattern to be transferred or placed on a silicon wafer or the like and the pattern of the reticle is projected on a reduced scale on the wafer through a projection lens. As a result, that which can be printed by a single exposure is a square area of the order of 10.times.10 mm on the wafer at the most. Therefore, in order to print a circuit pattern on the entire surface of a wafer of about 125 mm in diameter, a method is used which repeats the operation of moving a wafer-supporting stage a predetermined distance two dimensionally and then exposing the wafer (this method is hereinafter referred to as a step-and-repeat method). As a result, this type of exposure apparatus is sometimes known as a stepper.
Also, in the production of LSI circuits, more than several layers of patterns are registered and printed and in this case the registration error between the layers must be kept below a given value, that is, a registration error of as small as about 0.2 .mu.m, for example, is tolerable in the case of patterns having a minimum line-width of 1 .mu.m. Among the causes for such registration errors due to the exposure apparatus are .circle.1 a relative positional misplacement between the projected image and the wafer and .circle.2 a projection magnification error and projection distortion. To attain a highly accurate alignment for the purpose of reducing the registration error due to cause .circle.1 , two methods, i.e., an on-axis method, and off-axis method have heretofore been in use.
Of these methods, the on-axis method is also called the through-the-lens method (hereinafter referred to as a TTL), since the alignment marks of both a reticle and a wafer are observed simultaneously through a projection lens. On the contrary, the off-axis method is one which involves no such observation through a projection lens.
In the conventional reduction projection-type exposure apparatus, the step-and-repeat operation is performed in such a manner that a wafer is placed on a stage through a wafer holder and the stage is moved on the basis of the origin of rectangular coordinates forming a plane of movement of the stage. A reticle containing a circuit pattern is positioned and held above the wafer in such a way that the projected image of the reticle has as small a rotational displacement as possible with respect to the rectangular coordinates. In this condition, a series of exposure and printing operations by the step-and-repeat method, that is, the operation of determining a coordinate position on the rectangular coordinates, moving and holding the stage in position and effecting the exposure, is performed on repeatedly. However, if there is any shift in the detection center of a position sensor of a photoelectric microscope used in the alignment of the reticle, the reticle aligned on the basis of the position sensor involves an alignment error with respect to the rectangular coordinate system and therefore the resulting pattern area printed on the wafer generally tends to cause a rotational error.
On the other hand, the projection lens is required to have a resolution of the submicron order and it is designed so as to minimize its chromatic aberration with respect to the wavelength of the light used for printing purposes. As a result, if a light which is different in wavelength from the light for printing purposes (the printing light) is used for sensing the alignment marks, the focusing position tends to be shifted considerably due to a high chromatic aberration. As an example, consider a case where the wavelength of the printing light is 436 mm and the alignment mark sensing light has a wavelength of 633 mm. In order to cause the two lights to form image on the same plane on the wafer side, the reticle must be moved in the direction of the optical by an amount which is about 200 times the depth of focus of the projection lens. Then, in order to simultaneously observe the reticle and the wafer by the TTL method, it is most preferable to make the observation by using a light having a wave-length which is the same or close to that of the printing light. However, this observation method is disadvantageous in that the height of the irregularities in the circuit patterns formed on the wafer by the previous steps exceed the depth of focus of the projected image, in that the resolution condition is deteriorated by the interference effect between the printing light directed to the wafer and the reflected light from the wafer surface and so on.
Thus, a method is conceivable which comprises forming a first thick layer on the surface of a wafer to smooth the surface, forming on the first layer a second layer containing coloring matter for absorbing a printing light and further applying a photoresist to form a third layer on the second layer. However, this method gives rise to another disadvantage that if a light having the same or a similar wavelength as the wavelength of the printing light is used for alignment observing purposes or the like, the coloring matter in the second layer make it difficult to observe the objects (e.g., the marks provided for aligning purposes) below the second layer and hence it is impossible to ensure accurate alignment. On the other hand, while the absorption by the coloring matter in the second layer can be prevented if the observation is made by the TTL method using a light which is different in wavelength from the printing light, it, as such, is still impossible to ensure the desired accurate alignment due to the previously mentioned high chromatic aberration of the projection lens.
One of the causes in .circle.2 for the registration error, i.e., the distortion, is the distortion aberration of the projection optical system. The other of the causes in .circle.2 for the registration error, i.e., the magnification error, can be reduced by varying the distance between the reticle and the principal point (the principal plane) of the projection optical system if the reticle side light beam of the projection optical system is not telecentric and it can be reduced by relatively shifting in the direction of the optical axis the positions of the component elements inside the projection optical system (e.g., the optical members such as the lenses) if the light beam is telecentric. As a result, the error due to the causes in .circle.2 is constant if there is no change in the distance between the reticle and the projection lens and the relative positions of the component elements within the projection optical system and it can be considered as a systematic error.
Then, the error due to .circle.2 or the systematic error can be stably maintained at a small value over a long period of time if the apparatus is adjusted so as to maintain the value of the error below a predetermined value while measuring the value and it is also necessary to minimize the value during the adjustment in the manufacture of exposure apparatus. To measure the error due to .circle.2 , a method is conceivable in which an image of a reticle containing a pattern of marks at a plurality of predetermined positions, i.e., so-called test reticle, is printed in the photoresist on a wafer and the coordinates of the printed resist image of the marks are detected, thereby comparing the detected coordinates and the mark coordinates on the reticle. With this method, however, certain labor and time are required for exposing the pattern of the test reticle onto the wafer and developing the pattern and it is also necessary to use an expensive detecting apparatus for detecting the resist images of the marks.